[patch 0/2] Get watchdog reboot working on efika smartbook

[patch 0/2] Get watchdog reboot working on efika smartbook

[Arnaud Patard (Rtp)]

Hi,

With current patches in Sascha Hauer's tree, the efika smartbook is powering
off when rebooting. The cause of this bug is the default behaviour of the
mc13892 pmic. This patchset is configuring the pmic to reboot the system 
as expected by imx system.c (the reboot is done throught watchdog) and enables
the watchdog on the efika mx/sb systems.

Arnaud

[patch 1/2] mc13892: reboot on wdi event

[Arnaud Patard (Rtp)]

An embedded and charset-unspecified text was scrubbed...
Name: mc13892_wdog_reboot.patch
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20110228/562257ca/attachment.ksh>

[patch 1/2] mc13892: reboot on wdi event

[Mark Brown]

On Mon, Feb 28, 2011 at 02:21:33PM +0100, Arnaud Patard wrote:
> By default, on wdi (watchdog input) event the mc13892 is powering off.
> This patch is changing this to reboot on wdi event.
> 
> Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>

This strikes me as something that should be configurable by platforms.

[patch 1/2] mc13892: reboot on wdi event

[Arnaud Patard (Rtp)]

Mark Brown <broonie@opensource.wolfsonmicro.com> writes:

> On Mon, Feb 28, 2011 at 02:21:33PM +0100, Arnaud Patard wrote:
>> By default, on wdi (watchdog input) event the mc13892 is powering off.
>> This patch is changing this to reboot on wdi event.
>> 
>> Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
>
> This strikes me as something that should be configurable by platforms.

afaik, mc13892 is used only on imx designs and the reset for imx is done
only throught the watchdog. From this point of view,  I don't see any
valid reason for powering off instead of rebooting on watchdog event


Arnaud

[patch 1/2] mc13892: reboot on wdi event

[Mark Brown]

On Mon, Feb 28, 2011 at 04:25:45PM +0100, Arnaud Patard wrote:

> afaik, mc13892 is used only on imx designs and the reset for imx is done
> only throught the watchdog. From this point of view,  I don't see any
> valid reason for powering off instead of rebooting on watchdog event

Is there a separate way to do power off?

[patch 1/2] mc13892: reboot on wdi event

[Arnaud Patard (Rtp)]

Mark Brown <broonie@opensource.wolfsonmicro.com> writes:

> On Mon, Feb 28, 2011 at 04:25:45PM +0100, Arnaud Patard wrote:
>
>> afaik, mc13892 is used only on imx designs and the reset for imx is done
>> only throught the watchdog. From this point of view,  I don't see any
>> valid reason for powering off instead of rebooting on watchdog event
>
> Is there a separate way to do power off?

yeah. You can either disable pwgt1 and pwgt2 regulators or set the
useroff bit of the power control 0 register.

Arnaud

[patch 1/2] mc13892: reboot on wdi event

[Liam Girdwood]

On Mon, 2011-02-28 at 16:25 +0100, Arnaud Patard wrote:
> Mark Brown <broonie@opensource.wolfsonmicro.com> writes:
> 
> > On Mon, Feb 28, 2011 at 02:21:33PM +0100, Arnaud Patard wrote:
> >> By default, on wdi (watchdog input) event the mc13892 is powering off.
> >> This patch is changing this to reboot on wdi event.
> >> 
> >> Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
> >
> > This strikes me as something that should be configurable by platforms.
> 
> afaik, mc13892 is used only on imx designs and the reset for imx is done
> only throught the watchdog. From this point of view,  I don't see any
> valid reason for powering off instead of rebooting on watchdog event
> 

If this change is for i.MX only then it should be optional as Freescale
probably will sell this PMIC to non i.MX users.

Thanks

Liam

[patch 1/2] mc13892: reboot on wdi event

[Mark Brown]

On Mon, Feb 28, 2011 at 02:21:33PM +0100, Arnaud Patard wrote:

> +
> +	/* allows to reboot on wdi event */
> +	ret = mc13xxx_reg_read(mc13892, MC13892_POWERCTL2, &val);
> +	if (!ret) {
> +		val |= MC13892_POWERCTL2_WDIRESET;
> +		mc13xxx_reg_write(mc13892, MC13892_POWERCTL2, val);
> +	}
> +

Also, why is this in the regulator driver?  The MFD core for the device
might be a better choice - that'll ensure it gets done even if regulator
support isn't enabled for some reason.

[patch 1/2] mc13892: reboot on wdi event

[Arnaud Patard (Rtp)]

Mark Brown <broonie@opensource.wolfsonmicro.com> writes:

> On Mon, Feb 28, 2011 at 02:21:33PM +0100, Arnaud Patard wrote:
>
>> +
>> +	/* allows to reboot on wdi event */
>> +	ret = mc13xxx_reg_read(mc13892, MC13892_POWERCTL2, &val);
>> +	if (!ret) {
>> +		val |= MC13892_POWERCTL2_WDIRESET;
>> +		mc13xxx_reg_write(mc13892, MC13892_POWERCTL2, val);
>> +	}
>> +
>
> Also, why is this in the regulator driver?  The MFD core for the device
> might be a better choice - that'll ensure it gets done even if regulator
> support isn't enabled for some reason.

I was not sure where to put it so I put it in the place where the power
management is done. I can move it to the mfd driver. I'll only have to
make sure it's done only for mc13892 (I don't know what would be the
consequences to do it on mc13783 too and if there's a similar
configuration setting).

Arnaud

[patch 2/2] efika mx/sb: enable watchdog

[Arnaud Patard (Rtp)]

An embedded and charset-unspecified text was scrubbed...
Name: wdt.patch
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20110228/0e4de70b/attachment.ksh>

[patch 2/2] efika mx/sb: enable watchdog

[Uwe Kleine-König]

Hi Arnaud,

just one nitpick below ...

On Mon, Feb 28, 2011 at 02:21:34PM +0100, Arnaud Patard wrote:
> Register watchdog for efika mx and sb systems.
> 
> Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
> 
> Index: linux-2.6-submit/arch/arm/mach-mx5/mx51_efika.c
> ===================================================================
> --- linux-2.6-submit.orig/arch/arm/mach-mx5/mx51_efika.c	2011-02-27 23:13:37.000000000 +0100
> +++ linux-2.6-submit/arch/arm/mach-mx5/mx51_efika.c	2011-02-27 23:20:38.000000000 +0100
> @@ -611,6 +611,7 @@
>  					ARRAY_SIZE(mx51efika_pads));
>  	imx51_add_imx_uart(0, &uart_pdata);
>  	mx51_efika_usb();
> +	imx51_add_imx2_wdt(0, NULL);
>  	imx51_add_sdhci_esdhc_imx(0, NULL);
>  
>  	/* FIXME: comes from original code. check this. */
> Index: linux-2.6-submit/arch/arm/mach-mx5/Kconfig
> ===================================================================
> --- linux-2.6-submit.orig/arch/arm/mach-mx5/Kconfig	2011-02-27 23:13:42.000000000 +0100
> +++ linux-2.6-submit/arch/arm/mach-mx5/Kconfig	2011-02-27 23:20:38.000000000 +0100
> @@ -117,6 +117,7 @@
>  	bool
>  	select SOC_IMX51
>  	select IMX_HAVE_PLATFORM_IMX_UART
> +	select IMX_HAVE_PLATFORM_IMX2_WDT
>  	select IMX_HAVE_PLATFORM_SDHCI_ESDHC_IMX
>  	select IMX_HAVE_PLATFORM_SPI_IMX
These IMX_HAVE_PLATFORM entries are sorted alphabetically and 2 sorts
before _, so IMX_HAVE_PLATFORM_IMX2_WDT should be the first entry here.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |